Cowley, Benjamin R, Calhoun, Adam J, Rangarajan, Nivedita, Turner, Maxwell H, Pillow, Jonathan W, Murthy, Mala (October 2023) One-to-one mapping between deep network units and real neurons uncovers a visual population code for social behavior. bioRxiv. (Public Dataset) (Submitted)
Preview |
PDF
10.1101.2022.07.18.500505.pdf - Submitted Version Available under License Creative Commons Attribution Non-commercial No Derivatives. Download (36MB) | Preview |
Abstract
The rich variety of behaviors observed in animals arises through the complex interplay between sensory processing and motor control. To understand these sensorimotor transformations, it is useful to build models that predict not only neural responses to sensory input [1, 2, 3, 4, 5] but also how each neuron causally contributes to behavior [6, 7]. Here we demonstrate a novel modeling approach to identify a one-to-one mapping between internal units in a deep neural network and real neurons by predicting the behavioral changes arising from systematic perturbations of more than a dozen neuron types. A key ingredient we introduce is “knockout training”, which involves perturb-ing the network during training to match the perturbations of the real neurons during behavioral experiments. We apply this approach to model the sensorimotor transformation of Drosophila melanogaster males during a com-plex, visually-guided social behavior [8, 9, 10]. The visual projection neurons at the interface between the eye and brain form a set of discrete channels, suggesting each channel encodes a single visual feature [11, 12, 13]. Our model reaches a different conclusion: The visual projection neurons form a highly distributed population code that collectively sculpts social behavior. Overall, our framework consolidates behavioral effects elicited from various neural perturbations into a single, unified model, providing a detailed map from stimulus to neuron to behavior.
Actions (login required)
 |
Administrator's edit/view item |